Accuracy of magnetic resonance imaging in differentiating between benign and malignant vertebral lesions: Role of Diffusion-weighted imaging, in-phase/opposed-phase imaging and apparent diffusion coefficient

Objective: To determine the ability of MRI to distinguish between benign and malignant vertebral lesions. Material and methods: We included 85 patients and studied a total of 213 vertebrae (both pathologic and normal). For each vertebra, we determined whether the lesion was hypointense in T1-weighted sequences and whether it was hyperintense in STIR and in diffusion weighted sequences. We calculated the in-phase/out-of-phase quotient and the apparent diffusion coefficient for each vertebra. We combined parameters from T1-weighted, diffusion weighted, and STIR sequences to devise a formula to distinguish benign from malignant lesions. Results: The group comprised 60 (70.6%) women and 25 (29.4%) men with a mean age of 67 +/- 13.5 years (range, 33-90 y). Of the 85 patients, 26 (30.6%) had a known primary tumor. When the lesion was hypointense on T1-weighted sequences, hyperintense on STIR and diffusion-weighted sequences, and had a signal intensity quotient greater than 0.8, the sensitivity was 97.2%, the specificity was 90%, and the diagnostic accuracy was 91.2%. If the patient had a known primary tumor, these values increased to 97.2%, 99.4%, and 99%, respectively. Conclusion: Benign lesions can be distinguished from malignant lesions if we combine the information from T1-weighted, STIR, and diffusion-weighted sequences together with the in-phase/out-of-phase quotient of the lesion detected in the vertebral body on MRI. (C) 2013 SERAM. Published by Elsevier Espana, S.L.U. All rights reserved.